Glossary
Predictive analytics
Using historical data to forecast outcomes, demand, churn, risk, failure.
A credible forecast reports an interval, not a point, and is backtested against held-out historical periods.
Always compare against a naive baseline. Plenty of sophisticated models fail to beat 'same as last week', and that is worth knowing before deployment.
Adoption depends on how the forecast is presented. Planners who see a single number will either accept it blindly or override it entirely; planners who see a range, the main drivers and where the model is least confident will use it as an input to judgement, which is what it should be.
Decide in advance what result would make you stop. A forecast project with no threshold for abandonment tends to continue indefinitely on the strength of nearly-there results, and agreeing that number early is what prevents it.
Commonly misunderstood: Accuracy on historical data is not evidence about the future if the underlying process has changed.
Related terms, in context
The concepts you almost always meet alongside predictive analytics.
- Model drift
- Degradation over time as the world stops resembling the training data.
- Feature engineering
- Constructing the input variables a model learns from, still where most predictive performance comes from.
Where this shows up in our work
Predictive analytics is not an abstraction for us. It is a decision we make on live projects. It shows up most directly in predictive analytics & forecasting, where getting it wrong has a cost someone can measure.
If you are evaluating a vendor on this, the useful question is not whether they can define the term. It is what they measure, what they would refuse to do, and what happens in their system when the assumption behind predictive analytics stops holding.
Questions
What is Predictive analytics?
Using historical data to forecast outcomes, demand, churn, risk, failure.
What do people get wrong about predictive analytics?
Accuracy on historical data is not evidence about the future if the underlying process has changed.
Does Orqent Labs build this?
Yes, Predictive Analytics & Forecasting. We work across India, covering all 19,238 PIN codes remotely.
Building something that involves predictive analytics?
We will tell you honestly whether it is the right approach for your problem.
Or email bd@dtrasglobal.com · call +91 74118 77878